Writings on various topics (mostly technical) from Oliver Hookins and Angela Collins. We have lived in Berlin since 2009, have two kids, and have far too little time to really justify having a blog.
One of my favourite pastimes (sad as it is) is finding new and cool things that the Linux kernel can do. Short of reading through the actual source code, a relatively sane way to find your way around the kernel is by reading the documentation provided by the Linux Cross Reference. Seriously, bookmark it now.
Ever since my baptism of fire with our Sun OpenStorage cluster shortly after joining Nokia, I've actually quite admired the data security aspects of ZFS and RAIDZ. Former colleagues of mine at Anchor Systems are probably laughing at me now, as they've been running it for years. If you are running fairly regular scrubs on your pools there is very little chance of losing data due to random bits being corrupted on faulty drives. The software takes care of finding out when your hardware is likely to fail.
Wouldn't it be wonderful if Linux could do this as well with its md RAID driver? It turns out that it can, actually. If you run a simple command such as:
echo "check" > /sys/block/md0/md/sync_action
you will trigger a check of the consistency of your software RAID set. I have no idea yet if it is as reliable as ZFS scrub, but it seems like a worthwhile thing to schedule to run once a week. You can then check the contents of mismatch_cnt to see if any errors were found. Let's hope it outputs on which device the inconsistencies were found somewhere useful...